How to set up a weekly AI coaching summary agent for sales team leads in Demodesk

What and why

This guide shows you how to build a custom AI agent in Demodesk's AI Crew that delivers a weekly coaching summary to your sales team leads — automatically, every Friday. Team leads stop digging through individual call recordings and scorecards. They get one consolidated view of how the week's discovery calls went, where reps did well, and where the coaching gaps are.

Who this is for

Sales managers, team leads, and revenue operations owners who want consistent weekly coaching visibility across their AE team without manually compiling reports.

Prerequisites

• Demodesk Coaching & AI seat with AI Crew enabled
• AI Coach configured with at least one active scorecard, so there's call-level scoring data for the agent to summarize
• A delivery channel ready: email distribution list, Slack workspace, or Microsoft Teams channel
• A consistent meeting type or naming convention to filter on (in the example below: meetings titled “erstes Kennenlernen”)
• Admin or team lead permissions to create AI Crew agents

Steps

1. Open AI Crew from the Home sidebar — In the Demodesk app, go to Home and open the AI Crew section in the left sidebar, then create a new agent.
2. Choose “Build a Custom Agent”— Select the option to create your own agent rather than installing a pre-built one from the marketplace. This launches the AI Agent Builder.
3. Name the agent— Give it a clear, descriptive name like “Weekly Coaching Summary — Team Leads”. Team leads will see this name in their delivery channel, so make it self-explanatory.
4. Paste your custom prompt— In the prompt field, describe what the agent should do in plain language. A working example:

   Every Friday, summarize all “erstes Kennenlernen” meetings from the past 7 days. For each AE, include: number of calls held, average scorecard score, top 2 strengths, top 2 coaching gaps, and any deals flagged at risk. Group the output by AE. Keep it under 500 words and use bullet points.

   The AI Agent Builder turns this prompt into a configured agent. Refine it after the first test run.
5. Filter by meeting type— Add a filter so the agent only looks at the right calls. In this example, filter for meetings whose title contains “erstes Kennenlernen”. You can also filter by meeting type, host, team, or scorecard.
6. Connect the AI Coach data source— Make sure the agent has access to scorecard scores and coaching feedback, not just transcripts. Without this connection, the summary reads like a generic meeting recap instead of a coaching report.
7. Set the weekly schedule— Configure the agent to run weekly, every Friday at 9:00 AM (or whatever fits your team's cadence). A Friday morning run gives team leads the full week's calls and time to act on the summary before the weekend.
8. Choose the delivery channel— Select where the summary lands: email (to a team lead distribution list), Slack (to a private leads channel), or Microsoft Teams. Slack and Teams work well when team leads want to react and discuss in-thread. Email works better when the summary needs to be archived or forwarded to senior management.
9. Test with 1–2 team leads first— Before rolling out to the full leadership group, send the first 2–3 runs to one or two team leads. Review the output together. Tighten the prompt if the summary is too long, too vague, or missing the coaching angles you care about. This is the step most teams skip — and the one that determines whether the agent gets read or ignored.
10. Roll out to all team leads— Once the format is dialed in, add the rest of the team leads to the delivery channel and let the agent run.

Tips

• Run it manually before scheduling.Trigger a one-off run against last week's data to validate the output before committing to a recurring schedule.
• Keep the prompt specific to coaching, not deal status. If you want deal updates, build a separate agent. Mixing coaching feedback and pipeline updates in one summary dilutes both.
• Pair it with a per-rep agent. This agent goes to team leads. A second agent sends each AE their own weekly coaching feedback, so reps see their gaps in real time instead of hearing about them in 1:1s.
• Review the prompt quarterly.Sales motions change. The metrics that mattered last quarter aren't always the ones to coach on this quarter.
• Bookmark calls referenced in the summary. If retention is set to delete recordings after 30 days, bookmarking the calls the agent flags keeps them available for coaching reviews later.

Related skills and agents

• AI Coach— the underlying scorecard engine this agent reads from. Configure scorecards under Agents → Scorecards.
• AI CRM Concierge— for keeping HubSpot, Salesforce, or Pipedrive fields updated after every call. Separate from the coaching summary use case.
• Marketplace — browse pre-built coaching and reporting agents at marketplace.demodesk.ai/agents before building from scratch.